Cacao must be supported by the appropriate technology to increase
production. The cocoa seeds are one of the cocoa sections that can be utilized; they
must be dried prior to processing. However, during the grinding process, producers
only consider the seed's weight, ignoring the degree of drought uniformity. Given the
size variance of cocoa seeds, not all seeds will dehydrate uniformly during processing.
Using the CNN (Convolutional Neural Network) method and the YOLO (You Only
Look Once) architecture, the research concentrates on imaging for dry cocoa seed
detection based on color and shape using the CNN (Convolutional Neural Network)
method. With a total of 2880 images in the data set, the selection phase of cocoa seeds
commences by dividing the data set into two classes. The initial category is sun-dried
cacao. The final category is unrefined Cocoa seed. This study can distinguish between
dried and unprocessed cocoa seeds using multiple datasets with an average accuracy
of 99,8%, an average precision value of 99,15%, and an average recall of 99,8%.